Abstract T P36: Classification of Hemorrhagic Transformation by ECASS II Criteria More Severe on MRI Than CT

Abstract

Background and Purpose: ECASS II criteria for hemorrhagic transformation (HT) are based on CT, and no similar scoring system has been applied to MRI. Validation of MRI HT classification has implications for future MRI-based clinical trials and for clinical management. The purpose of our study was to apply and compare HT classification using T2*-weighted MRI and CT. Methods: We reviewed the scans of consecutive stroke patients treated with IV thrombolytic therapy from 2002 to 2012 who had post-treatment scans with CT bracketed by MRIs within 72h of CT, or CT and MRI within 24h of each other. Separately for both CT and MRI, the same 2 independent readers blinded to clinical information assigned HT classification using ECASS II criteria. A third reader adjudicated discrepancies. Inter-rater and inter-modality reliability were determined. For the inter-modality comparison, final readings were included for CT-MRI pairs obtained within 3 hours and cases with consistent MRI reads obtained before and after CT. Results: Of the 121 patients included, 79 had bracketed MRIs with CT and 42 had a CT-MRI pair (CT before MRI in all cases). Inter-rater agreement was similar for CT and MRI HT classifications between the 2 raters (CT: 80% agreement, Cohen’s κ=0.692, n=119; MRI: 85% agreement, Cohen’s κ=0.739, n=187). Between CT and MRI there was good agreement across the 5 ECASS HT categories (82%, Cohen’s κ=0.65), and for combined hemorrhagic infarction (HI) vs. parenchymal hematoma (PH) (88%; Cohen’s κ=0.754). For those patients identified as having HT on either modality (n=24), HT was rated as more severe using MRI compared to CT (p=0.026) (Figure). Conclusions: ECASS II classification of HT by MRI has good inter-rater and inter-modality agreement overall compared to CT, but for those with hemorrhage, HT is rated as more severe using MRI. Therefore, MRI classification of HT should not be equated to that of CT.